Article Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/IC
A Tungsten-Based Nanolaminated Ternary Carbide: (W,Ti)4C4−x Babak Anasori,*,†,‡ Jun Lu,*,§ Oleg Rivin,∥ Martin Dahlqvist,§ Joseph Halim,§ Cooper Voigt,† Johanna Rosen,§ Lars Hultman,§ Michel W. Barsoum,*,† and El’ad N. Caspi†,∥ †
Department of Materials Science and Engineering, Drexel University, Philadelphia, Pennsylvania 19104, United States A. J. Drexel Nanomaterials Institute, Drexel University, Philadelphia, Pennsylvania 19104, United States § Thin Film Physics Division, Department of Physics, Chemistry and Biology (IFM), Linköping University, SE-581 83 Linköping, Sweden ∥ Physics Department, Nuclear Research Centre − Negev, 84190 Beer-Sheva, Israel Inorg. Chem. Downloaded from pubs.acs.org by NEW MEXICO STATE UNIV on 01/05/19. For personal use only.
‡
S Supporting Information *
ABSTRACT: Nanolamellar transition metal carbides are gaining increasing interests because of the recent developments of their twodimensional (2D) derivatives and promising performance for a variety of applications from energy storage, catalysis to transparent conductive coatings, and medicine. To develop more novel 2D materials, new nanolaminated structures are needed. Here we report on a tungstenbased nanolaminated ternary phase, (W,Ti)4C4−x, synthesized by an Alcatalyzed reaction of W, Ti, and C powders at 1600 °C for 4 h, under flowing argon. X-ray and neutron diffraction, along with Z-contrast scanning transmission electron microscopy, were used to determine the atomic structure, ordering, and occupancies. This phase has a layered hexagonal structure (P63/mmc) with lattice parameters, a = 3.00880(7) Å, and c = 19.5633(6) Å and a nominal chemistry of (W,Ti)4C4−x (actual chemistry, W2.1(1)Ti1.6(1)C2.6(1)). The structure is comprised of layers of pure W that are also twin planes with two adjacent atomic layers of mixed W and Ti, on either side. The use of Al as a catalyst for synthesizing otherwise difficult to make phases, could in turn lead to the discovery of a large family of nonstoichiometric ternary transition metal carbides, synthesized at relatively low temperatures and shorter times.
■
INTRODUCTION
sites, are interleaved with pure A-layers. Non-MAX phase ternary structures are known as well.8,9 The discovery in 2011 that Al-containing MAX phases could be readily converted to their 2D counterparts, labeled MXenes, by simply immersing in fluoride containing acidic solutions for a few hours at ambient temperature, has given the MAX field, especially Al-containing ones, a large boost in terms of interest and potential applications, not as MAX phases per se, but as precursors to MXenes.10,11 A more in depth discussion on MXenes is presented in the Supporting Information. Examples of recently discovered MAX phases include quaternaries with two M elements that are either ordered out-of-plane12−16 or inplane,17−19 ternary phases with double A layers,20,21 or Aelement layers substituted with noble transition metals.22 Despite several theoretical predictions of W-containing MAX phases, only two W-containing MAX phases have been discovered to date: the in-plane ordered (W2/3Sc1/3)2AlC and (W2/3Y1/3)2AlC phases.19 The dearth of compounds combining W and Ti is not only limited to the MAX phases. There are only few studies on TMCs comprised of W and Ti. The first carbide containing Ti and W, was reported in the late of 1930s, but without detailed structure information.23 A cubic (Ti,W)C
Group IV to VI transition metal carbides (TMCs) and nitrides (TMNs) are sometimes referred to as metallic ceramics, because they combine high metallic conductivities with ceramic properties such as chemical stability, high hardness, and high melting points. Some are used as cutting tools, wearresistant, and high-temperature materials.1 Several nanostructure designs of this family of materials have been explored after their superior catalytic behavior was demonstrated.2,3 Additionally, group III to VI transition metals form ternary nanolaminated structures by combining with A-group elements. The Mn+1AXn or MAX phases (where M is an early transition metal, A is an A-group element, mostly groups 13 and 14 of the periodic table, and X is C and/or N with n = 1 to 3) are one such example. The first MAX phase, Cr2AlC, was reported by Bloom and Grant in 1950, and labeled an Hphase.4 In 1960, Kudielka and Rohde discovered a second phase, Ti2SC.5 In the 1960s, Nowotny et al. discovered over 50 such compounds.6 Barsoum et al., however, were the first to make predominantly single phase bulk MAX phase materials and document their unusual set of properties, including machinability.7 The MAX phases are hexagonal (space group P63/mmc) with two formula units per unit cell, in which closepacked M atoms, with interstitial X atoms in the octahedral © XXXX American Chemical Society
Received: August 6, 2018
A
DOI: 10.1021/acs.inorgchem.8b02226 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
Figure 1. SEM of as synthesized (W,Ti)4C4‑x powder. (a−c) Secondary electron SEM micrographs showing elongated, plate-like grains with similar morphologies to the nanolayered structure of the MAX phases. Few grains with different morphology at the bottom of (a) are impurities. (b,c) Higher-magnification images of platelike grains again show the layered structure. Kinks are marked by arrows in (c) and shown in inset. (d) Backscattered electron SEM on a pressed pellet of as-synthesized powder showing at least three regions. The majority of the grains, some of which are labeled A, are bright gray and flaky. EDX of these grains showed only the presence of W, Ti, and C with a W/Ti atomic ratio of 2:1.1 ± 0.1. The dark gray grains (labeled B) are most probably WAl4 and the black grains (labeled C) are Al2O3. The latter two are impurities found in the assynthesized powder.
with rock salt structure was first reported by Nowotny et al. in 1954.24 Rock salt solid solutions of (Ti1−x,Wx)C with different W contents were also synthesized by hot pressing TiC and WC mixtures at ∼2600 °C.25 In another study, cubic single crystals of (Ti1−x,Wx)C were grown from an Al melt at 1600 °C, in which only 20 mol % of WC was accommodated in the (Ti,W) C crystals.26 Crucially, no hexagonal ordered layered structures of W, Ti and C has been reported to date.27−29 Here, we report on the first hexagonal close packed (hcp) W-containing ternary phase with a M4C4−x chemistry starting with a W/Ti/Al/C powder mixture. The original intent was to produce an out-of-plane ordered MAX phase. Somewhat surprisingly, the resulting phase with a nominal chemistry of (W,Ti)4C4−x and actual chemistry of W2.1(1)Ti1.6(1)C2.6(1) was Al free. This phase is layered, ordered, and quite different from the structure of either hexagonal WC,30 or cubic TiC.31 The resulting powder was characterized using a scanning electron microscope (SEM), X-ray diffraction (XRD), selected electron diffraction (SAED), energy dispersive spectrum (EDX), highresolution scanning transmission electron microscope (HRSTEM), and neutron powder diffraction (NPD).
■
idea, we mixed W/Ti/C powders with no Al in the stoichiometric ratio of 2:1:2 and followed the same procedure outlined above. In this case, the resulting material crystallized in a mixture of TiCx and W2C (Supporting Information, Figure S1). A SEM (Supra50, Zeiss) equipped with an EDX spectroscope (EDX, Oxford Inca X-Sight) was used to qualitatively characterize the powder morphology and its purity. To improve the quality of the backscattered electron SEM micrographs, the powders were cold pressed into a disk at a load corresponding to a stress of 100 MPa. To determine the composition of individual grains, we performed EDX on individual grains by zooming on a single particle (magnifications of 50 to 100 k×) and ran EDX scans for ∼7 s. XRD was carried out on a Rigaku SmartLab (Japan) diffractometer using step scans of Δ 2θ = 0.02° in the 2θ range of 5−120°, and a dwell time of 7 s per step. A Cu Kα source was used with voltage and current of 40 kV and 44 mA, respectively. The accuracy of the diffractometer in determining lattice parameters, and its instrumental peak-shape function parameters were found using LaB6 (NIST 660B) standards resulting in a systematic shift in lattice parameters determination of ∼−0.06%. The NPD measurement was performed at room temperature, RT, (∼300 K) at the KANDI-II powder double-axis diffractometer in the Israel Research Reactor No. 2 (IRR-2) at the Nuclear Research Centre, Negev, Israel (Israel Atomic Energy Commission). The diffractometer and its 10 3He-based neutron multidetector, were calibrated using various standard samples (e.g., Si, Al2O3). The incident neutron beam [λ = 2.47(1)] was obtained from the (002) reflection of a pyrolytic graphite monochromator. A ∼7g sample was loaded into vanadium, V, cylindrical sample holder with a 10 mm diameter and ∼0.2 mm wall thickness. Both the NPD and XRD diffraction patterns were analyzed by the Rietveld analysis32 method using the FULLPROF code.33 HRSTEM and EDX were also carried out with a double-corrected FEI Titan3 60−300 operated at 300 kV, equipped with the Super-X EDX system. SAED characterization was performed using a FEI Tecnai G2 TF20 UT instrument equipped with a field emission gun that runs at a voltage of 200 kV and a point resolution of 0.19 nm. The specimens were prepared by embedding the powder in a Ti grid, reducing the Ti-grid thickness down to 50 μm via mechanical polishing and finally Ar+ ion milling to reach electron transparency.
EXPERIMENTAL DETAILS
The sample was prepared by ball milling elemental powders of W, Ti, Al, C (all from Alfa Aesar) with particle sizes 1.4.34 Backscattered electron SEM micrographs with EDX taken on the pressed discs (Figure 1d) showed the presence of three distinct regions; a majority of W, Ti, C phase (bright gray; labeled A) where EDX showed a W/Ti atomic ratio of 2:1.1(1), and impurity phases, WAl4 (dark gray; B) and Al2O3 (black; labeled C). The SAED on the as-synthesized powder (Figure 2a,b) can be indexed with a P63/mmc hexagonal structure. EDX compositional analysis in the HRTEM on different spots revealed that this compound is indeed Al free at the precision level of our EDX instrument (Figure 2c). The preliminary atomic coordinates of W and Ti were obtained from analytical HRSTEM. The Z-contrast images along the [112̅0] and [101̅0] zone axes shown in Figure 2d,e, respectively, clearly show bright and gray spots, arranged in separate layers. Since there is no Al, and C is too light to be visible in such images, we only discuss the positions of the W and Ti atoms. Three bright atomic layers and one gray layer are clearly seen, suggesting a W-rich/Ti-rich ordered layered structure. A more detailed intensity profile (inset in Figure 2d,e) clearly demonstrates that the middle bright layer has stronger scattering than the adjacent two less bright layers, indicating either variations in the W/Ti relative occupancy or the existence of vacancies in the different layers. Inspection of the local composition viewed along the [112̅0] direction reveals partitioning of W (red) and Ti (green) with
three W-rich layers and one Ti-rich layer (see Figure 2f−h). Moreover, the EDX line scan along the [0001] direction (Figure 2h) shows a strong Ti signal along the gray layer together with a weak W signal, suggesting the presence of a Tirich layer. The brightest layers appear to be comprised of W alone. The other two adjacent layers show a strong secondary W signal together with weaker Ti peaks. These observations agree with the HRSTEM intensity profile (inset in Figure 2d,e) and support a partially ordered solid solution nanolayered structure that can be expressed as (Ti1−zWz)(W1−wTiw)W(W 1−w Ti w )(Ti 1−z W z )(W 1−w Ti w )W(W 1−w Ti w ). Note that throughout this paper, the Ti atoms are colored green, the W atoms, red. The enlarged Z-contrast image along the [112̅0] zone axis (Figure 3a) illustrates the stacking sequence and the atomic positions of the M elements, in which three different contrasts are apparent. Position 2d at z = 0.75 (brightest layer, labeled h in Figure 3b) is occupied by W only. Position 2a at z = 0, gray layer, is occupied mainly by Ti with some W referred to as (Ti,W) in Figure 3b. The two adjacent (W,Ti) layers, described by the 4f positions, with a z along the c-axis of ∼0.125, are referred to as (W,Ti) in Figure 3b. If one ignores the C atoms, then the atomic arrangement,shown by capital letters on left of Figure 3b, is CBABCABACBAB···, where the underlined layers are pure W. In other words, the W-only planes are twin mirror planes. In that respect, they are similar to the A-layers in the MAX phases. It is easier to understand the atomic ordering by using the Wycoff-Jagodzinski notation, also known as the chnotation.35,36 In this notation, a close-packed plane is marked as h, if its two adjacent planes are symmetrically located relatively to it, similar to a hcp structure (ABABAB). However, if the adjacent planes are nonsymmetrically located relative to the given atomic plane, the plane is marked by c, similar to a cubic close-packed (ABCABC). The (W,Ti)4C4−x unit cell can thus be best succinctly described as hc3 (cchccchcc) as shown in blue letters on right of Figure 3b. Although hc3 ordering was predicted before,35,36 to our knowledge, it has previously never C
DOI: 10.1021/acs.inorgchem.8b02226 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
to refine such a model, that is (W,Ti)3AlC2 with either ordered W/Ti layers or disordered ones,38 resulted in extremely poor agreement between model and experiment for all cases (not shown). However, assuming the hexagonal space group P63/ mmc for an Al-free phase of (W,Ti)4C4−x (x ≈ 1.4) with cell parameters, a ≈ 3.08 Å and c ≈ 20.47 Å, obtained from the HRSTEM images (Figure 3) are in good agreement with those obtained from the XRD patterns, viz. a ∼ 3.01 Å, and c ∼ 19.56 Å. It is worth repeating here that to further test the role of Al, we sintered a similar stoichiometric ratio of W/Ti/C powders with no Al under the same conditions. The resulting material crystallized in a mixture of TiCx and W2C (Supporting Information, Figure S1), and (W,Ti)4C4−x was not formed. Since the synthesis conditions (1600 °C, 4 h) chosen here are typical of several Al-containing MAX phases, such as Ta4AlC3, Mo2TiAlC2, and Mo2Ti2AlC3.11−13 it is quite unlikely that the Al simply evaporated away in the tube furnace. Since a large number of occupancy parameters had to be determined, a simultaneous Rietveld analysis of both the XRD and NPD results was carried out. Using this method enables individual refinements of each element occupancy for each crystallographic site. The model included the major hexagonal (W,Ti)4C4−x, cubic carbide (W,Ti)C, and Al2O3 phases. On the basis of the STEM result, the starting (W,Ti)4C4−x structure was assumed to have the hexagonal P63/mmc space group with the following atomic positions: (W,Ti)1 in 2d(1/ 3,2/3,3/4), (W,Ti)2 in 4f(1/3,2/3, z ∼ 1/8), (W,Ti)3 in 2a(0,0,0). For the carbon positions we assumed C1 in 4e(0,0, z ∼ 1/3), and C2 in 4f(1/3,2/3, z ∼ 0), covering all the octahedral sites between the M layers. The starting occupancies used were a ratio of 2:1 for W/Ti, and half occupancy for C1 and full occupancy for C2. For the (W,Ti)C minor impurity phase, we used the same cubic Fm3̅m space group of TiC39 with a cell parameter of ∼4.29 Å. According to ref 25, this cell parameter corresponds to (W0.4Ti0.6)C, which served as a starting value for the relative Ti/W occupancy of
Figure 3. Atomic ordering of (W,Ti)4C4−x. (a) HRSTEM image viewed along the [112̅0] direction, showing atomic planes with different contrast. The brightest layer is W, the two adjacent layers are W-rich (W,Ti) layers. The gray layers with lowest brightness are Tirich or (Ti,W) layers. (b) Structure model and stacking sequence of (W,Ti)4C4−x, where W, Ti, and C are shown by red, green, and black spheres, respectively. The ABC stacking notation is shown in black, on the left side of the structure and the Wycoff-Jagodzinski notation (ch) is shown in blue on the right side demonstrating hc3 ordering.
been experimentally observed. In our HRSTEM microscopy, we also observed a stacking fault (Figure S2a), where a Ti-rich layer is replaced with a W-rich layer (see colored arrows in Figure S2b). The XRD patterns of the acid-cleaned and water-washed powders are shown in Figure 4a. Peaks belonging to three phases are identified. One set of reflections belong to the major phase and can be attributed to a hexagonal P63/mmc symmetry with lattice parameters a and c of ∼3.01 and ∼19.56 Å, respectively. Another set belong to a minor phase with Fm3̅m cubic symmetry with a lattice parameter a ∼ 4.298 Å (Figure 4a). A 1−3 wt % impurity Al2O3 phase is also present. Although the symmetry and lattice parameters are reminiscent to what is expected from a MAX phase with n = 2,37 an attempt
Figure 4. Crystal structure of (W, Ti)4C4−x. (a,b) Joint Rietveld analysis of (a) XRD, and (b) NPD data (red crosses). The refined model (solid black line) is comprised of three phases: hexagonal P63/mmc that describes the major phase with the refined formula of W2.1(1)Ti1.6(1)C2.6(1) (first row of tags), cubic Fm3̅m that describes the minor (W,Ti)C phase (second row of tags), and a minor Al2O3 impurity phase (third row of tags). The quality of the fit is represented by the difference between model and observed data (blue lines at the bottom). (c) Suggested structure for W2.1(1)Ti1.6(1)C2.6(1), with random C and W vacancies and W, Ti intermixing in different planes. D
DOI: 10.1021/acs.inorgchem.8b02226 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry this minority phase. Elemental occupancies were refined with no constraints for the major hexagonal phase and minor cubic phase. In addition, all models allowed for the refinement of the phases’ cell parameters, isotropic thermal displacements for all crystallographic sites in (W,Ti)4C4−x and their counterparts in the cubic phase, atomic positions, phase fractions, linear interpolation type background,33 Lorentzian-type peak broadening for all phases, and instrument asymmetry parameters. Excellent fit between the same structural model and the observed XRD (Figure 4a), and NPD (Figure 4b) data was achieved. Table 1 summarizes the refined values obtained from the Rietveld refinements. According to our refinement, the structure consists of one Ti-rich layer, presented as the (W,Ti)3 sites in Table 1, two Wrich layers on the (W,Ti)2 sites and one pure W layer with ∼25% vacancies, presented as the (W,Ti)1 sites in Table 1. The occupancy of C and atomic positions were also accurately determined with an overall ∼35% of disordered C vacancies per formula unit for (W,Ti)4C4−x. The unit cell is shown in Figure 4c, which as noted above has a hc3 stacking ordering. The twin mirror h plane is pure W with ∼25% vacancies and the two immediate adjacent layers are almost equimolar in W and Ti, viz. W0.574(2)Ti0.40(4). The second adjacent layer to the mirror plane is Ti-rich, with almost a 1:3 W/Ti ratio, viz. W0.25(1)Ti0.78(4). The C atoms occupy the interstitial sites and their total per formula unit is estimated to be 2.6. On the basis of our NPD results, there is no evidence for ordering of the C vacancies and all the interstitial layers are partly occupied by C atoms. It is important to note that since all the interstitial sites are partially filled with C, the (W,Ti)4C4−x formula with ∼35% C vacancies (x = 1.4) is the correct notation. However, if future work shows evidence for C-ordering, then the correct notation is (W,Ti)4C3, in which it is implicitly assumed that at some point the C vacancies would order.35,36,40 If such ordering is not forthcoming, then the correct notation is (W,Ti)4C4 phase as assumed in this study. Further assessment is needed to identify the best notation. For the minority (W,Ti)C cubic phase, the refined W/Ti ratio was 0.44(5)/0.65(7) in agreement with the previous results,25 adding to the reliability of the refinement process. Combining EDX and the XRD and NPD refinement results, we conclude that chemistry of this carbide is most probably W2.1(1)Ti1.6(1)C2.6(1). Note that the uncertainties in the occupancies determined by the refinement process are statistical in nature. The final formula is reported with larger uncertainties covering the systematic uncertainty as well. In our previous work, we argued that in the out-of-plane ordered MAX phases, the M elements, whose binary carbides did not crystallize in the rock salt structure avoided the center of the MX blocks.13,41 The same principle appears to apply here. The pure W layers are sitting in an ABA, or hexagonal, arrangement. It is only when these layers are alloyed with Ti, that they are no longer sitting in a hexagonal arrangement, but rather in a face-centered cubic (fcc), ABC, type arrangement. This is consistent with the fact that it is only when the Ti/W ratio >1 that the rock salt carbide structures become stable.25 Note that in the layers adjacent to the mirror planes the W/Ti ratio is only slightly over 1:1.
Table 1. Structural Parameters Determined from Simultaneous Rietveld Analysis of (W,Ti)4C4−x NPD and XRD Dataa phase
parameters
refined values
a (Å) c (Å) wt % wt % (W,Ti)1
3.00880(7) 19.5633(6) 77(5) 79(6) 0.25(8) 0.75(1) 0b 0.1275(2) 0.885(8) 0.574(2) 0.40(4) 1.3(2) 0.25(1) 0.78(4)d 0.3098(3) 0.425(4) −0.0640(2) 0.87(1) 6.75 14.2 4.2942(7) 19(4) 20(1) 0.44(5) 0.65(7) 0.84(1) 2.83 0.307 7.58 7.17 10.4 7.12
(W,Ti)2
(W,Ti)4C4−x
(W,Ti)3
C1c C2c RF a (Å) wt % (W,Ti)C
Rexp Rwp
W Ti Bov RF
NPD XRD Biso (Å2) W occ. Ti occ. z Biso (Å2) W occ. Ti occ. Biso (Å2) W occ. Ti occ. z occ. z occ. NPD XRD NPD XRD occ. occ. NPD XRD NPD XRD NPD XRD
a
The refined formula is W2.1(1)Ti1.6(1)C2.6(1) which crystallizes in the hexagonal P63/mmc space-group with (W,Ti)1 positioned in 2d(1/ 3,2/3,3/4), (W,Ti)2 in 4f(1/3,2/3,z ∼ 1/8), (W,Ti)3 in 2a(0,0,0), C1 in 4e(0,0,z ∼ 1/3), and C2 in 4f(1/3,2/3,z ∼ 0). The minor (W,Ti)C phase orders in the cubic Fm3̅m space-group with (W,Ti) positioned in 4a(0,0,0), and C in 4b(1/2,1/2,1/2); a and c are cell parameters; wt % is a phase’s weight percent, Biso (Bov) is individual (overall) atomic thermal displacement parameter; occ. is occupancy; RF is individual phase agreement factor; Rexp is overall expected agreement factor, and Rwp is the weighted profile agreement factor. Note that χ2 = (Rwp/Rexp)2.33 Numbers in parentheses represent one standard deviation of the last significant digits determined by the refinement process. A small Al2O3 impurity (
